OBJECTIVE--To evaluate the incidence of intraoperative and early postoperative complications (up to two months after implant) of endocardial permanent pacemaker insertion in all patients under-going a first implant at a referral centre. METHODS--Prospective evaluation of all endocardial pacemaker implantation procedures performed from April 1992 to January 1994 carried out by completion of standard audit form at implant. Patients' demographic data, medical history, details of pacemaker hardware used, and any complications were noted. Follow up information was also collected prospectively onto standard forms at pacemaker outpatient clinic. SETTING--United Kingdom tertiary referral cardiothoracic centre. PATIENTS--1088 consecutive patients underwent implantation of their first endocardial permanent pacemaker from April 1992 to January 1994. Implant and follow up data were available for 1059 (97.3%) patients at analysis. The median (range) age was 77 years (16-99); 51.2 % were male. RESULTS--Dual chamber units were implanted in 54.1% of patients, single chamber atrial in 5.2%, and ventricular in 40.7%. A temporary pacing lead was present at implant in 22.9% of patients. Most (93.6%) implants were performed via the subclavian vein. Immediate complications were rare: eight (0.8%) patients developed pneumothorax requiring medical treatment and 11 (1.0%) an insignificant pneumothorax. There was no significant difference in the pneumothorax rate for dual chamber (DDD) compared with single chamber systems. Arterial puncture without sequelae was documented in 2.7% of attempts at subclavian vein cannulation. A total of 35 patients (3.3%) required reoperation; the reoperation rate for dual chamber (3.5%) was similar to that for single chamber (3.1%) systems. Electrode displacement (n = 15, 1.4%) was the most common reason for reoperation. Atrial lead displacement (n = 10, 1.6% of atrial leads) was significantly more common than ventricular lead displacement (n = 5, 0.5% of ventricular leads, P = 0.047). There was no difference in electrode displacement rates for dual (1.6%) compared with single (1.2%) chamber systems. Pacemaker pocket infection led to reoperation in 10 patients (six dual, four single chamber, P = not significant) and was significantly more common in patients who had a temporary pacing lead in place at implant (2.9%) than in those who did not (0.4%, P = 0.0014). Five patients (0.5%) required reoperation for generator erosion (two dual, three single chamber, P = not significant). and a further five for drainage of haematoma or a serous fluid collection (three dual, two single chamber, P = not significant). Complications that did not require reoperation were also rare. Undersensing occurred in 10 patients (0.9%). Atrial undersensing (n = 8) was significantly more common than ventricular undersensing (n = 2, P = 0.017). All patients were successfully treated by reprogramming of sensitivity. Superficial wound infection was treated successfully with antibiotics in nine patients (six dual, three single chamber, P = not significant). Three patients with DDD generators developed sustained atrial fibrillation: two required reprogramming to VVI mode and one required cardioversion. CONCLUSIONS--Permanent pacing in a large tertiary referral centre with experienced operators carries a low risk. Infection rates are low, < 1% overall but significantly higher in patients who undergo temporary pacing before implantation. Lead displacement and undersensing are more likely to occur with atrial than ventricular leads. The overall complication rate for dual chamber pacing, however, is no higher than for single chamber pacing.